Arsenic metabolism has typically been studied by administering arsenate or arsenite into animals and humans and then studying the metabolites excreted in the urine. Although such studies have yielded information about the beginning and the end of the metabolic pathways for the metabolism of inorganic arsenic compounds, any statements as to the molecular mechanisms of these reactions have had to be highly speculative. Now that the rabbit and the rhesus monkey liver enzymes that transfer methyl groups from S-adenosylmethionine to arsenite and monomethlyarsonic acid have been purified and the reactions characterized, meaningful investigations of species diversity and polymorphism of these enzymes have become possible. New World animals studied thus far appear to be deficient in or totally lacking these enzymes. Old World animals, with the exception of the chimpanzee, have ample amounts of arsenite and monomethylarsonic acid methyltransferases. A hypothesis that the lack of arsenite methyltransferases may have had an evolutionary advantage for certain species is proposed.